Physicians' Preferences for Hospice If They Were Terminally Ill and The

Letters

in 5.6% and 0.36%, respectively, only marginally less than the 6. Vakil N, Moayyedi P, Fennerty MB, Talley NJ. Limited value of alarm features prevalence in those with symptomatic GERD.7 in the diagnosis of upper gastrointestinal malignancy: systematic review and meta-analysis. Gastroenterology. 2006;131(2):390-401. The difficulty in identifying the right patients with GERD 7. Rex DK, Cummings OW, Shaw M, et al. Screening for Barrett’s esophagus in to endoscope is demonstrated in the study by Kramer et al,8 colonoscopy patients with and without heartburn. Gastroenterology. published in this issue. In an impressive audit of nearly 500 000 2003;125(6):1670-1677. cases of GERD identified in Veteran Health Administration rec- 8. Kramer JR, Shakhatreh MH, Naik AD, Duan Z, El-Serag HB. Use and yield of ords, the authors found that those patients more likely to re- endoscopy in patients with uncomplicated gastroesophageal reflux disorder ceive an EGD in the first year after diagnosis were actually less [published online January 27, 2014]. JAMA Intern Med. doi:10.1001/ jamainternmed.2013.12756. likely to have either BE or esophageal cancer.8 Men older than 9. Dulai GS, Guha S, Kahn KL, Gornbein J, Weinstein WM. Preoperative 65 years, for example, were nearly 7-fold more likely to have prevalence of Barrett’s esophagus in esophageal adenocarcinoma: a systematic BE or esophageal cancer compared with a cohort of women review. Gastroenterology. 2002;122(1):26-33. younger than 50 years as the reference group, but these men 10. Chang JY, Locke GR III, McNally MA, et al. Impact of functional were also significantly less likely to undergo EGD (odds ratio, gastrointestinal disorders on survival in the community. Am J Gastroenterol. 0.77).8 2010;105(4):822-832. We still miss most patients with BE despite the wide- spread use of EGD; up to 95% of cases of adenocarcinoma occur in the setting of no prior diagnosis of BE.9 The data from Kramer et al8 suggest that this may in part be due to Physicians’ Preferences for Hospice if They Were underutilization of EGD in high-risk cases. On the other Terminally Ill and the Timing of Hospice Discussions hand, the mortality of symptomatic reflux is not increased With Their Patients over the background population,10 and there is still no con- Physicians often delay hospice discussions with their termi- vincing evidence that endoscopic screening of symptomatic nally ill patients despite guidelines recommending such dis- GERD will reduce esophageal adenocarcinoma rates (stop- cussions for patients expected to die within 1 year,1,2 but ping smoking and losing weight would probably be more reasons for this are not well understood. Evidence suggests valuable).3 Until noninvasive biomarkers to identify BE are that physicians “practice what they preach” when counsel- available, judicious application of EGD to those with alarm ing about health behaviors,3 although their treatment rec- signals or other high-risk cases (eg, those with chronic ommendations may not necessarily reflect their own prefer- heartburn, who are older white men, or those with chronic ences, with one study suggesting they recommend more reflux who are obese) seems reasonable but is unlikely to conservative treatments than they might choose for save many lives. themselves.4 As physicians may prefer less aggressive end- of-life care than their patients generally receive,5 physicians’ personal preferences for hospice may influence their Nicholas J. Talley, MD, PhD, FRACP approach to hospice discussions with their terminally ill Kate E. Napthali, FRACP patients. We examined physicians’ reported preferences for hos- Author Affiliations: University of Newcastle, Newcastle, New South Wales, pice enrollment if they were terminally ill. We also assessed Australia (Talley); John Hunter Hospital, Newcastle, New South Wales, Australia (Napthali); Mayo Clinic, Rochester, Minnesota (Talley). whether physicians who would enroll in hospice if terminally Corresponding Author: Nicholas J. Talley, MD, PhD, FRACP, University of ill differed from others in the timing of hospice discussions with Newcastle, Callaghan, NSW 2308, Australia ([email protected]). their patients. Published Online: January 27, 2014. doi:10.1001/jamainternmed.2013.12992. Methods | This study was approved by the institutional review Conflict of Interest Disclosures: None reported. boards at all participating institutions. We surveyed physi- 1. Evans JA, Early DS, Fukami N, et al; ASGE Standards of Practice Committee; cians caring for patients with cancer enrolled in the multire- Standards of Practice Committee of the American Society for Gastrointestinal gional population-based Cancer Care Outcomes Research and Endoscopy. The role of endoscopy in Barrett’s esophagus and other 1 premalignant conditions of the esophagus. Gastrointest Endosc. Surveillance (CanCORS) study. Informed consent was im- 2012;76(6):1087-1094. plied by physicians’ participation in the survey. Physicians in- 2. Solaymani-Dodaran M, Logan RF, West J, Card T, Coupland C. Risk of dicated on a 5-point Likert scale how strongly they agreed or oesophageal cancer in Barrett’s oesophagus and gastro-oesophageal reflux. disagreed with the statement “If I were terminally ill with can- Gut. 2004;53(8):1070-1074. cer, I would enroll in hospice.” They were also asked to as- 3. Kahrilas PJ, Shaheen NJ, Vaezi MF; American Gastroenterological Association sume that they were caring for an asymptomatic patient with Institute; Clinical Practice and Quality Management Committee. American Gastroenterological Association Institute technical review on the management advanced cancer, who they believed had 4 to 6 months to live of gastroesophageal reflux disease. Gastroenterology. 2008;135(4):1392-1413. and report whether they would discuss hospice with the pa- 4. Shaheen NJ, Weinberg DS, Denberg TD, Chou R, Qaseem A, Shekelle P; tient “now,” “when the patient first develops symptoms,” Clinical Guidelines Committee of the American College of Physicians. Upper “when there are no more nonpalliative treatments to offer,” endoscopy for gastroesophageal reflux disease: best practice advice from the “only if the patient is admitted to the hospital,”or “only if the clinical guidelines committee of the American College of Physicians. Ann Intern 1 Med. 2012;157(11):808-816. patient and/or family bring it up.” 5. Katz PO, Gerson LB, Vela MF. Guidelines for the diagnosis and management Among 4488 respondents (response rate 61%), we ex- of gastroesophageal reflux disease. Am J Gastroenterol. 2013;108(3):308-328. cluded 105 who did not answer the hospice self-preference

466 JAMA Internal Medicine March 2014 Volume 174, Number 3 jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/25/2021 Letters

Table. Physician Willingness to Enroll in Hospice: Personal and Practice Characteristicsa

Unadjusted Proportion Who Strongly Agree They Would Enroll in Hospice if Strongly Agreeing They Terminally Ill With Cancer, Would Enroll in Hospice, Physician Characteristics Total, No. (%) % Adjusted OR (95% CI)b Total 4368 (100) 64.5 Age, y ≤39 959 (22) 65.4 1 [Reference] 40-49 1264 (29) 65.5 1.08 (0.89-1.31) 50-54 767 (18) 67.1 1.18 (0.95-1.47) 55-59 702 (16) 65.4 1.20 (0.96-1.50) ≥60 676 (15) 57.7 1.00 (0.79-1.25)

Sex Abbreviation: OR, odds ratio. Male 3453 (80) 61.9 1 [Reference] a Percentages include only reported, Female 837 (20) 76.1 1.80 (1.49-2.18) nonimputed values and may not sum to 100% because of rounding Specialty or missing values. Missing values Primary care physician 1743 (41) 69.5 1 [Reference] were present for the following Surgery 923 (22) 56.6 0.65 (0.55-0.78) variables: sex (n = 78), specialty (n = 77), proportion of patients in Medical oncology 600 (14) 70.3 0.93 (0.74-1.17) managed care (n = 374), and Radiation oncology 257 (6) 57.6 0.57 (0.42-0.76) number of terminally ill patients Other specialty 768 (18) 61.2 0.75 (0.62-0.90) cared for in the past year (n = 62). Adjusted analyses used imputed Patients in managed care, % data. ≤50 2330 (58) 60.8 1 [Reference] b Adjusting for all variables in the ≥51 1664 (42) 69.8 1.30 (1.12-1.51) table, as well as type of practice. No. of terminally-ill patients Board certification, US medical in the last year school graduate status, and level of ≤12 2308 (54) 62.5 1 [Reference] teaching involvement were not associated in adjusted analyses and ≥13 1998 (46) 67.1 1.29 (1.12-1.50) were not included in the model.

hospice “now” with their terminally ill patients. We omitted Figure. Physician Willingness to Enroll in Hospice and Report of Early Hospice Discussions With Terminally Ill Patients With Cancer variables with adjusted P values >.10.

35 Results | Respondents’ characteristics are given in the Table. Most respondents strongly (64.5%) or somewhat (21.4%) 30 agreed they would enroll in hospice if terminally ill. In 25 adjusted analyses, physicians who were female, cared for more terminally ill patients, and worked in managed-care 20 settings were more likely than others to strongly agree they 15 would enroll in hospice. Surgeons and radiation oncologists were less likely than primary care physicians to strongly 10

Discuss Hospice “Now,” % Discuss Hospice “Now,” agree they would enroll in hospice. Adjusted Proportion Who Would 5 Overall, 26.5% reported they would discuss hospice “now” with a patient who had 4 to 6 months to live. Other 0 Strongly agree Other responses physicians reported they would wait until the patient has Personal Preferences for Hospice symptoms (16.4%), there were no more treatments to offer Enrollment if Terminally Ill (48.7%), the patient and/or family brings it up (4.3%), or the patient is hospitalized (4.1%). After adjustment, physicians question and 15 likely trainees who graduated after 2004. Mul- who strongly agreed they would enroll in hospice them- tiple imputation was used to address item nonresponse in the selves were more likely than other physicians to report dis- adjusted analyses.6 cussing hospice “now” (odds ratio, 1.7; 95% CI, 1.5-2.0) We used multivariable logistic regression to examine (Figure). physician and practice factors associated with physicians’ strong agreement that they would enroll in hospice if termi- Discussion | Most physicians reported they would enroll in hos- nally ill with cancer. In a second model, we assessed if phy- pice if they were terminally ill with cancer, particularly women, sicians who strongly agreed they would enroll in hospice primary care physicians, and those in managed-care settings were more likely than other physicians to report discussing and with more terminally ill patients. Physicians with strong

jamainternalmedicine.com JAMA Internal Medicine March 2014 Volume 174, Number 3 467

Downloaded From: https://jamanetwork.com/ on 09/25/2021 Letters

personal preferences for hospice were more likely than oth- 1. Keating NL, Landrum MB, Rogers SO Jr, et al. Physician factors associated ers to report discussing hospice with their patients earlier. Phy- with discussions about end-of-life care. Cancer. 2010;116(4):998-1006. sicians should consider their personal preferences for hos- 2. National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: palliative care: version 1. 2013. http://www.nccn.org/professionals pice as a factor as they care for terminally ill patients with /physician_gls/pdf/palliative.pdf. Accessed May 21, 2013. cancer. Physicians with negative views of hospice may con- 3. Wells KB, Lewis CE, Leake B, Ware JE Jr. Do physicians preach what they sider pursuing additional education about how hospice may practice? A study of physicians’ health habits and counseling practices. JAMA. help their patients. 1984;252(20):2846-2848. 4. Ubel PA, Angott AM, Zikmund-Fisher BJ. Physicians recommend different Garrett M. Chinn, MD, MS treatments for patients than they would choose for themselves. Arch Intern Med. 2011;171(7):630-634. Pang-Hsiang Liu, MD, PhD 5. Gramelspacher GP, Zhou X-H, Hanna MP, Tierney WM. Preferences of Carrie N. Klabunde, PhD, MHS, MBA physicians and their patients for end-of-life care. J Gen Intern Med. Katherine L. Kahn, MD 1997;12(6):346-351. Nancy L. Keating, MD, MPH 6. He Y, Zaslavsky AM, Landrum MB, Harrington DP, Catalano P. Multiple imputation in a large-scale complex survey: a practical guide. Stat Methods Med Res. 2010;19(6):653-670. Author Affiliations: Division of General Medicine, Department of Medicine, Massachusetts General Hospital, and Harvard Medical School, Boston (Chinn); Department of Health Care Policy, Harvard Medical School, Boston, Massachusetts (Liu, Keating); Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, Maryland (Klabunde); RAND, Posttraumatic Stress Disorder and Santa Monica, California (Kahn); Division of General Internal Medicine, Medication Nonadherence in Patients With Department of Medicine, David Geffen School of Medicine, Los Angeles, Uncontrolled Hypertension California (Kahn); Division of General Internal Medicine and Primary Care, Department of Medicine, Brigham and Women's Hospital, Boston, Posttraumatic stress disorder (PTSD) is common in primary care 1 Massachusetts (Keating). patients and is associated with psychological distress, sui- Corresponding Author: Nancy L. Keating, MD, MPH, Department of Health cide risk, and disability. Posttraumatic stress disorder also in- Care Policy, Harvard Medical School, 180 Longwood Ave, Boston, MA creases risk of incident and recurrent cardiovascular events,2 02115-5899 ([email protected]). possibly by reducing medication adherence.3 Prior studies Published Online: December 16, 2013. showing an association between PTSD and medication doi:10.1001/jamainternmed.2013.12825. nonadherence3 are limited by their use of self-report to mea- Author Contributions: Drs Liu and Keating had full access to all of the data in sure adherence as PTSD can bias reporting of negative the study and take responsibility for the integrity of the data and the accuracy behaviors.4 We evaluated the association between PTSD and of the data analysis. Study concept and design: Chinn, Liu, Keating. antihypertensive medication adherence using electronic moni- Acquisition of data: Klabunde, Kahn, Keating. toring in primary care patients with uncontrolled hyperten- Analysis and interpretation of data: Chinn, Liu, Klabunde, Kahn, Keating. sion. Drafting of the manuscript: Chinn, Liu, Keating. Critical revision of the manuscript for important intellectual content: Chinn, Liu, Klabunde, Kahn, Keating. Methods | The institutional review board of Columbia Univer- Statistical analysis: Chinn, Liu, Keating. sity Medical Center, New York, New York, approved the pro- Obtained funding: Kahn, Keating. tocol. All patients provided written informed consent. We en- Administrative, technical, or material support: Klabunde. Study supervision: Kahn, Keating. rolled a convenience sample of patients with uncontrolled hypertension from an academic hospital-based primary care Conflict of Interest Disclosures: None reported. Funding/Support: This work of the Cancer Care Outcomes Research and Surveillance (CanCORS) Consortium was supported by grants from the National Figure. Posttraumatic Stress Disorder (PTSD) Symptoms and Cancer Institute (NCI) to the Statistical Coordinating Center (U01 CA093344) Nonadherence to Antihypertensive Medications and the NCI-supported Primary Data Collection and Research Centers (Dana Farber Cancer Institute/Cancer Research Network [U01 CA093332], Harvard 100 Medical School/Northern California Cancer Center [U01 CA093324], RAND/UCLA [U01 CA093348], University of Alabama at Birmingham [U01 CA093329], University of Iowa [U01 CA093339], University of North Carolina 80 [U01 CA093326]) and by a Department of Veterans Affairs grant to the Durham VA Medical Center [CRS 02-164]. Dr Keating’s effort was also funded by grant 60 1R01CA164021-01A1 from the NCI. Role of the Sponsors: Dr Klabunde is an employee of the NCI. Aside from her 40 contributions, the funding agencies had no role in design and conduct of the

study; in the collection, analysis, and interpretation of the data; or in the % of Nonadherence to

preparation, review, or approval of the manuscript Blood Pressure Medication 20 Previous Presentation: Portions of this work were presented in abstract form at the Society of General Internal Medicine’s 36th Annual Meeting; April 25, 0 2013; Denver, Colorado. 0 1-2 3-4 Additional Contributions: Robert Fletcher, MD, MSc, Harvard Medical School, PTSD Symptom Score provided helpful comments on an earlier version of the manuscript. Nonadherence was defined as taking less than 80% of the antihypertensive Correction: This article was corrected on January 8, 2014, to fix the value of the regimen. Error bars represent 95% CIs. number of respondents reported in the Methods section.

468 JAMA Internal Medicine March 2014 Volume 174, Number 3 jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/25/2021